CN110738188A - Ancient character recognition system based on presorting - Google Patents
Ancient character recognition system based on presorting Download PDFInfo
- Publication number
- CN110738188A CN110738188A CN201911015645.1A CN201911015645A CN110738188A CN 110738188 A CN110738188 A CN 110738188A CN 201911015645 A CN201911015645 A CN 201911015645A CN 110738188 A CN110738188 A CN 110738188A
- Authority
- CN
- China
- Prior art keywords
- ancient
- ancient character
- font
- character
- characters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/22—Character recognition characterised by the type of writing
- G06V30/226—Character recognition characterised by the type of writing of cursive writing
- G06V30/2264—Character recognition characterised by the type of writing of cursive writing using word shape
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The invention provides ancient character recognition systems capable of directly recognizing different types of ancient characters and guaranteeing recognition accuracy, which are characterized by comprising an ancient character font storage part, an ancient character image acquisition part, a font parameter analysis part, a category comparison division part and an ancient character recognition part, wherein the ancient character font storage part stores font information corresponding to different types of ancient characters, the ancient character image acquisition part is used for acquiring an ancient character image to be recognized, the font parameter analysis part is used for analyzing the font of the ancient characters in the ancient character image so as to obtain corresponding font parameters, the category comparison division part is used for comparing the font parameters with various types of font information respectively and dividing the ancient character image corresponding to the font parameters into corresponding ancient character categories, and the ancient character recognition part is used for completing the ancient character recognition on the ancient character image according to the ancient character categories.
Description
Technical Field
The invention belongs to the field of ancient character recognition, and particularly relates to ancient character recognition systems based on pre-classification.
Background
Ancient characters are ancient characters generated along with the change of history, and as for China, there are many characters belonging to different historical times, such as oracle characters, golden characters and the like, and for each types of ancient characters, researchers need to invest a lot of time to carry out intensive research.
In order to facilitate the research work of the ancient characters, the existing ancient character recognition technology can realize the recognition of the ancient characters through the modes of characteristic comparison, machine learning and the like of the ancient characters generally, so that a computer can automatically recognize corresponding modern characters according to the input ancient characters, and a user can conveniently know the content of the ancient characters.
However, because the fonts of various types of ancient characters have very large style differences, various ancient character recognition technologies only limit recognition objects within small batches of data, such as only gold character recognition or only ruin oracle character recognition, and are limited in software application.
Disclosure of Invention
In order to solve the problems, ancient character recognition systems which can directly recognize different types of ancient characters and can ensure the recognition accuracy are provided, the invention adopts the following technical scheme:
the invention provides an pre-classification-based ancient character recognition system which is used for effectively recognizing different types of ancient characters and is characterized by comprising an ancient character font storage part, an ancient character image acquisition part, a font parameter analysis part, a category comparison division part and an ancient character recognition part, wherein the ancient character font storage part is used for storing font information corresponding to the different types of ancient characters, the ancient character image acquisition part is used for acquiring an ancient character image to be recognized, the font parameter analysis part is used for analyzing the font of the ancient characters in the ancient character image so as to obtain corresponding font parameters, the category comparison division part is used for comparing the font parameters with the various types of font information respectively and dividing the ancient character image corresponding to the font parameters into corresponding ancient character types, and the ancient character recognition part is used for completing the ancient character recognition on the ancient character image according to the ancient character types.
The ancient character recognition system based on the presorting can also have the technical characteristics that the font information comprises font thickness ranges and font edge sharpness ranges of various ancient characters, the font parameters comprise font thickness parameters and font edge sharpness degrees of the ancient characters in the ancient character image, and the type comparison division part takes the type corresponding to the font information in accordance with the font parameters as the ancient character type when the font parameters are respectively compared with various types of font information.
The invention provides pre-classification-based ancient character recognition systems which are used for effectively recognizing different types of ancient characters and are characterized by comprising a classification model storage part, an ancient character image acquisition part, a character pattern parameter analysis part, a type recognition classification part and an ancient character recognition part, wherein the classification model storage part stores a type classification model which is trained in advance and used for recognizing the type of the ancient characters, the ancient character image acquisition part is used for acquiring the ancient character image to be recognized, the character pattern parameter analysis part is used for analyzing the character pattern of the ancient characters in the ancient character image so as to obtain corresponding character pattern parameters, the type recognition classification part is used for inputting the character pattern parameters into the type classification model for recognition and classifying the ancient character image corresponding to the character pattern parameters into corresponding ancient character types, and the ancient character recognition part is used for recognizing the ancient characters according to the ancient character types.
The ancient character recognition system based on pre-classification provided by the invention can also have the technical characteristics that the font parameter analysis part quantizes the font into the font parameters through the Fourier function.
The ancient character recognition system based on the pre-classification provided by the invention can also have the technical characteristics that: ancient character data set storage portion stores the ancient character data set that corresponds different ancient character class, and wherein, ancient character recognition portion includes: the recognition model storage unit is used for storing a plurality of ancient character recognition models which are obtained by training according to different types of ancient character data sets in advance by adopting a machine learning method; the model retrieval and acquisition unit is used for retrieving the recognition model storage unit according to the ancient characters and acquiring the ancient character recognition models of corresponding types as detection models; and the extraction and identification unit is used for inputting the ancient character image into the detection model to extract corresponding characteristics and finish ancient character identification.
The ancient character recognition system based on the pre-classification provided by the invention can also have the technical characteristics that: ancient character data set storage portion stores the ancient character data set that corresponds different ancient characters class respectively and contain ancient characters digital image and corresponding modern character information, and wherein, ancient characters recognition portion includes: a data set retrieval and acquisition unit, which is used for retrieving the ancient character data set storage part according to the ancient character class and acquiring the corresponding ancient character data set as a detection data set; the image comparison and acquisition unit is used for respectively comparing the ancient character image with the ancient character image in the detection data set and acquiring the most similar ancient character image in the detection data set as a comparison result image; and the retrieval and identification unit is used for retrieving the ancient character data set storage part according to the comparison result image and acquiring corresponding modern character information so as to finish ancient character identification.
Action and Effect of the invention
According to the ancient character recognition system based on presorting, the font of the ancient characters in the ancient character image is analyzed by the font parameter analysis part so as to quantize the font of the ancient characters into recognizable font parameters, the ancient character image is divided into corresponding ancient character classes according to the comparison of the font parameters and the font information by the class comparison division part, and the ancient character image is recognized by the ancient character recognition part according to the ancient character classes, so that the ancient character type to be recognized is recognized automatically, accurate ancient character recognition is completed according to the corresponding ancient character types.
In addition, the invention also provides ancient character recognition systems which finish the ancient character presorting by the machine learning method, because the ancient character digital image is divided into corresponding ancient character classes by adopting the category division model to recognize the font parameters through the category recognition division part after the font parameter analysis part analyzes the font parameters, the division can be more accurately finished when the ancient characters are presorted, meanwhile, the category division model can also optimize the classification effect through continuous iteration in steps, and finally, the ancient character recognition part can more accurately finish the recognition according to the corresponding ancient character classes.
Drawings
FIG. 1 is a block diagram of a pre-classification based ancient character recognition system according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a seal-printed character font according to an embodiment of the present invention;
FIG. 3 is a schematic view of a golden text (bronze inscription) glyph in an embodiment of the invention;
FIG. 4 is a schematic illustration of an oracle font of an embodiment of the present invention;
FIG. 5 is a schematic diagram of the "day" word of different ancient languages in the example of the present invention;
FIG. 6 is a flowchart of the ancient character recognition process in an embodiment of the present invention
Fig. 7 is a block diagram of a ancient character recognition system based on pre-classification according to a second embodiment of the present invention.
Detailed Description
In order to make the technical means, the creation features, the achievement purposes and the effects of the invention easy to understand, the ancient character recognition system based on the pre-classification of the invention is specifically described below with reference to the embodiment and the attached drawings.
< example >
FIG. 1 is a block diagram of a pre-classification based ancient character recognition system in an embodiment of the present invention.
As shown in fig. 1, the ancient character recognition system 100 based on presorting includes an ancient character data set storage unit 11, an ancient character font storage unit 12, an ancient character image acquisition unit 13, a font parameter analysis unit 14, a category comparison and classification unit 15, an ancient character recognition unit 16, a screen storage unit 17, an input display unit 18, a system-side communication unit 19, and a system-side control unit 20 that controls the above units.
The system-side communication unit 19 is used for data communication between the respective components of the ancient character recognition system 100 or between the ancient character recognition system 100 and another system. The system-side control unit 20 stores a computer program for controlling the operations of the respective components of the ancient character recognition system 100.
In this embodiment, the ancient character recognition system 100 is computers held by users, and the users can input the ancient character images into the computers or select the ancient character images stored in the computers, so that the system side control unit 20 controls the corresponding parts to complete the ancient character recognition according to the program.
The ancient character data set storage unit 11 stores ancient character data sets corresponding to different types.
In this embodiment, the ancient character data sets are of different types corresponding to different times, different carriers and different regions, such as royal, carapace, and golden languages, each types of ancient character data sets includes fixed number of ancient character pictures of the ancient characters and modern character information (such as modern characters, character codes, etc.) corresponding to the ancient characters.
The ancient character font storage unit 12 stores font information corresponding to ancient characters of different types.
In this embodiment, the font information is obtained by analyzing various ancient texts in advance, and includes information such as font stroke thickness ranges and font edge sharpness ranges corresponding to the various ancient texts.
The ancient character image acquisition part 13 is used for acquiring an ancient character image to be recognized.
In this embodiment, the ancient character image acquired by the ancient character image acquiring unit 13 is an ancient character image input or selected by the user.
The font parameter analyzing unit 14 is configured to analyze the font of the ancient character in the ancient character image to obtain a corresponding font parameter.
In this embodiment, the font parameters include font thickness parameters and font edge sharpness of the ancient texts in the ancient text image. The seal-printed characters shown in fig. 2, the golden texts (bronze inscriptions) shown in fig. 3, and the oracle-bone texts shown in fig. 4 have very different font styles, so that the font parameter analysis unit 14 can analyze and quantize the fonts of the fonts to obtain font parameters, and can identify the types of the ancient characters, thereby facilitating the computer to complete subsequent identification.
In this embodiment, the font parameter analyzing unit 14 performs recognition by a fourier function to obtain corresponding font parameters. As shown in fig. 5, taking the word "day" as an example: the different classes of ancient characters exhibit significant differences in the font stroke weight range and the font edge sharpness range. The oracle bone characters are fine in shape and stroke; the strokes of the Shanzhou golden character are thick and heavy, and the stippling is often presented in a very fat shape; spring and autumn warring country gold characters are lengthened; the warring country Chu simple characters are handwriting by ink, the strokes are uneven in thickness, and the tail strokes of partial strokes are sharp; the Qin and Xiao seal characters are symmetrical in strokes, and the tail ends of the strokes are mellow. The font characteristics are obvious, and after the font characteristics are identified by using the Fourier function, the relevant parameters have obvious differences.
The category comparison and classification section 15 is configured to compare the font parameters with the respective categories of font information stored in the ancient character font storage section 12, and classify the ancient character images corresponding to the font parameters into corresponding ancient character categories.
The type comparison and classification section 15 determines whether the font thickness parameter and the font edge sharpness in the font parameter conform to the font thickness range and the font edge sharpness range in the font information when classifying the ancient character type to which the ancient character image belongs, and uses the type corresponding to the font information conforming to the range as the ancient character type.
The ancient character recognition unit 16 is configured to perform ancient character recognition on the ancient character image based on the ancient character class classified by the class comparison classification unit 15.
In this embodiment, the ancient character recognition unit 16 completes ancient character recognition by machine learning, and the ancient character recognition unit 16 includes a recognition model storage unit 161a, a model retrieval obtaining unit 162a, and an extraction and recognition unit 163 a.
The recognition model storage unit 161a stores a plurality of ancient character recognition models corresponding to different types, which are obtained by performing training in advance according to different types of ancient character data sets by using a machine learning method.
In this embodiment, the process of training the ancient character recognition model in advance includes: firstly, a plurality of basic models are constructed through a conventional neural network technology, then iterative training is carried out according to various types of ancient character data sets stored in an ancient character data set storage part 11 until the models are converged, and finally, the trained models are correspondingly divided according to the types of the ancient character data sets during training so as to obtain ancient character recognition models corresponding to different types.
The model retrieval obtaining unit 162a is configured to retrieve the recognition model storage unit based on the ancient character class and obtain the ancient character recognition model of the corresponding class as the detection model.
The extraction and recognition unit 163a is used for inputting the ancient character image into the detection model to extract corresponding features and complete ancient character recognition.
In this embodiment, when the ancient character classes include a plurality of classes, the model retrieving and acquiring unit 162a acquires a plurality of corresponding ancient character recognition models as the detection models, and the extracting and recognizing unit 163a sequentially inputs the ancient character images into the detection models, outputs the modern character information as the recognition result if the modern character information recognized by the detection models is the same, outputs the modern character information in the same number if the modern character information recognized by the detection models is different, and outputs the reference character information in the largest number if the machine recognition still cannot lock the accurate modern character information.
The ancient character recognition unit 16 can also recognize the ancient characters by image matching, and in this case, the ancient character recognition unit 16 includes a dataset search acquisition unit 161b, an image matching acquisition unit 162b, and a search recognition unit 163 b.
The dataset retrieval acquiring unit 161b is configured to retrieve an ancient character dataset storage section 11 based on the ancient character class and acquire a corresponding ancient character dataset as a detection dataset.
The image comparison and acquisition unit 162b is configured to compare the ancient text digital image with the ancient text digital images in the detection data set, and acquire the ancient text digital image most similar in the detection data set as a comparison result image.
The retrieval and identification unit 163b is used for retrieving the ancient character data set storage part according to the comparison result image and acquiring the corresponding modern character information so as to complete ancient character identification.
In this embodiment, when the ancient character classes include a plurality of classes, the data set retrieving and acquiring unit 161b also acquires a plurality of corresponding ancient character data sets as the detection data sets, and the image comparison and acquiring unit 162b sequentially performs comparison to acquire the comparison result image, so that the retrieving and identifying unit 163b finally acquires a plurality of modern character information, and if the plurality of modern character information are the same, outputs the modern character information as the identification result, and if the modern character information identified by each detection model are different, outputs the most modern character information in the same number.
The screen storage unit 17 stores a result display screen.
The result display screen is used for displaying when the ancient character recognition part 16 finishes ancient character recognition and displaying an ancient character image and corresponding modern character information in the screen for the user to check.
The input display part 18 is used for displaying the picture, thereby leading the user to complete corresponding human-computer interaction. In the present embodiment, the input display unit 18 is an input/output device such as a computer monitor and a keyboard, and in other embodiments, the input display unit 18 may be another input/output device such as a touch panel.
FIG. 6 is a flow chart of an ancient character recognition process in an embodiment of the invention.
As shown in fig. 6, the process of ancient character recognition by the ancient character recognition system 100 comprises the following steps:
step S1, the ancient character image acquisition section acquires an ancient character image to be recognized, and then proceeds to step S2;
step S2, the font parameter analyzing unit 14 analyzes the ancient character image obtained in step S1 and obtains corresponding font parameters, and then the process goes to step S3;
step S3, the category comparison and classification section 15 compares the font parameters analyzed in step S2 with the font information stored in the ancient character font storage section 12 and classifies the ancient character images into corresponding ancient character categories, and then the process goes to step S4;
step S4, the ancient character recognition part 16 recognizes the ancient character image according to the ancient character class divided in step S3 and obtains the corresponding modern character information, and then the step S5 is performed;
in step S5, the input display unit 18 displays the result display screen and displays the ancient character image and the corresponding modern character information in the screen for the user to view, and then enters the end state.
< example two >
In the second embodiment, the same reference numerals are given to the components having the same configurations as those in embodiment , and the description thereof will be omitted.
In the second embodiment, ancient character recognition systems for pre-classifying ancient characters by machine learning method are provided, comparing with embodiment , as shown in fig. 7, the difference is that the ancient character font storage unit 12 and the category comparison division unit 15 are replaced with a division model storage unit 21 and a category recognition division unit 22, and the specific differences are as follows:
the classification model storage unit 21 stores a class ratio model that is trained in advance and used for identifying the class to which the ancient character belongs.
In this embodiment, the category classification model is constructed based on a conventional neural network model, training is completed through a large number of ancient character images marked with the categories to which the ancient character images belong, and category classification of the ancient characters can be performed after the model training is completed. In addition, the category division model can optimize the ancient character classification effect through continuous iteration.
The category identification and classification unit 22 is used for inputting the font parameters into a category classification model for identification and classifying the ancient character images corresponding to the font parameters into corresponding ancient character categories.
The process of identifying ancient characters in the second embodiment is substantially the same as that in embodiment , wherein after the font parameter analysis unit 14 analyzes the font parameters, the steps to be performed are different from step S3 in embodiment , specifically, after step S2 goes to step S3a, the category identification and classification unit 22 inputs the font parameters obtained by analyzing in step S2 into the category classification model stored in the classification model storage unit 21 and classifies the ancient character image into corresponding ancient character categories, and then goes to step S4. in addition, step S2 and the previous steps and step S4 and the subsequent steps in the second embodiment are the same as those in embodiment , and will not be described herein again.
Examples effects and effects
According to the ancient character recognition system based on presorting provided by the embodiment, because the font of the ancient characters in the ancient character image is analyzed by the font parameter analysis part, the font of the ancient characters is quantized into recognizable font parameters, the ancient character image is divided into corresponding ancient character classes according to the comparison of the font parameters and the font information by the class comparison division part, the ancient character image is recognized by the ancient character recognition part according to the ancient character classes, so that the ancient character recognition system can automatically recognize the types of the ancient characters to be recognized and accurately recognize the ancient characters according to the corresponding types.
In addition, this embodiment two still provides kinds of ancient character recognition systems that accomplish ancient character presorting through the machine learning method, because after font parameter analysis part analysis font parameter, thereby divide ancient characters digital image into corresponding ancient characters class through type identification division portion adoption type division model discernment to font parameter, so can accomplish the division more accurately when carrying out ancient characters presorting, simultaneously, thereby kind division model also can be through constantly iterating and optimizing classification effect step by step, finally make ancient characters recognition portion can accomplish the discernment according to corresponding ancient characters class more accurately.
In addition, in the embodiment , since the font parameters are compared with the parameter ranges in the font information and the corresponding ancient character categories are used as the ancient character categories, the category comparison and division unit can more completely divide the categories which the ancient character images may correspond to, thereby effectively reducing the classification errors of pre-classification caused by mistakes and omissions.
In addition, in the embodiment, because ancient character recognition department acquires corresponding ancient character recognition model through ancient character class and accomplishes the discernment of ancient characters, consequently can be in advance carry out the pertinence training to all kinds of ancient characters respectively with each ancient character recognition model, thereby make each ancient character recognition model can be effective, accurately carry out the ancient character recognition task under each kind separately, thereby it makes ancient character recognition system can accomplish accurate discernment to different kinds of ancient characters to go forward step through ancient character recognition department selection suitable model.
In addition, in the embodiment, the ancient character recognition part acquires the corresponding ancient character data set through the ancient characters to complete recognition of the ancient characters, so that the training, maintenance and other costs for the ancient character recognition model can be omitted, and the recognition task can be completed more simply under the condition of ensuring the ancient character recognition efficiency.
The above-described embodiments are merely illustrative of specific embodiments of the present invention, and the present invention is not limited to the description of the above-described embodiments.
Claims (6)
1, ancient character recognition system based on presorting, which is used for effectively recognizing different kinds of ancient characters, and is characterized by comprising:
an ancient character font storage part for storing font information corresponding to different kinds of ancient characters;
an ancient character image acquisition part for acquiring an ancient character image to be recognized;
a font parameter analysis part for analyzing the font of the ancient characters in the ancient character image to obtain corresponding font parameters;
a category comparison dividing part for comparing the font parameters with various types of font information and dividing the ancient character image corresponding to the font parameters into corresponding ancient character categories; and
and the ancient character recognition part is used for completing ancient character recognition on the ancient character image according to the ancient character class.
2. The system of claim 1, wherein the system further comprises:
wherein the font information comprises font thickness ranges and font edge sharpness ranges of various ancient characters,
the font parameters comprise font thickness parameters and font edge sharpness of the ancient characters in the ancient character image,
the type comparison and classification section may determine the type corresponding to the font information corresponding to the font parameter as the ancient character type when the font parameter is compared with the font information of each type.
ancient character recognition system based on presorting, which is used for effectively recognizing different kinds of ancient characters, and is characterized by comprising:
a classification model storage unit that stores a category classification model trained in advance and used for identifying a category to which the ancient character belongs;
an ancient character image acquisition part for acquiring an ancient character image to be recognized;
a font parameter analysis part for analyzing the font of the ancient characters in the ancient character image to obtain corresponding font parameters;
a category identification dividing part for inputting the font parameters into the category dividing model for identification and dividing the ancient character image corresponding to the font parameters into corresponding ancient character categories; and
and the ancient character recognition part is used for completing ancient character recognition on the ancient character image according to the ancient character class.
4. The system of claim 1 or 3, wherein:
wherein the glyph parameter analysis unit quantizes the glyph into a glyph parameter by a Fourier function.
5. The system of claim 1 or 3, further comprising:
an ancient character data set storage part for storing ancient character data sets corresponding to different ancient character classes,
wherein, ancient characters discernment portion includes:
the recognition model storage unit is used for storing a plurality of ancient character recognition models which are obtained by training according to the ancient character data sets of different types in advance by adopting a machine learning method;
the model retrieval and acquisition unit is used for retrieving the recognition model storage unit according to the ancient character classes and acquiring the ancient character recognition models of corresponding classes as detection models; and
and the extraction and identification unit is used for inputting the ancient character image into the detection model to extract corresponding characteristics and finish the ancient character identification.
6. The system of claim 1 or 3, further comprising:
an ancient character data set storage part for storing ancient character data sets which respectively correspond to different ancient character classes and contain ancient character digital images and corresponding modern character information,
wherein, ancient characters discernment portion includes:
a data set retrieval and acquisition unit, which is used for retrieving the ancient character data set storage part according to the ancient character class and acquiring the corresponding ancient character data set as a detection data set;
the image comparison and acquisition unit is used for respectively comparing the ancient character images with the ancient character images in the detection data set and acquiring the ancient character images which are most similar in the detection data set as comparison result images; and
and the retrieval and identification unit is used for retrieving the ancient character data set storage part according to the comparison result image and acquiring corresponding modern character information so as to finish the ancient character identification.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911015645.1A CN110738188A (en) | 2019-10-24 | 2019-10-24 | Ancient character recognition system based on presorting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911015645.1A CN110738188A (en) | 2019-10-24 | 2019-10-24 | Ancient character recognition system based on presorting |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110738188A true CN110738188A (en) | 2020-01-31 |
Family
ID=69271082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911015645.1A Pending CN110738188A (en) | 2019-10-24 | 2019-10-24 | Ancient character recognition system based on presorting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110738188A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104794455A (en) * | 2015-05-04 | 2015-07-22 | 北京信息科技大学 | Dongba hieroglyphic recognizing method |
| CN105335689A (en) * | 2014-08-06 | 2016-02-17 | 阿里巴巴集团控股有限公司 | Character recognition method and apparatus |
| CN107273897A (en) * | 2017-07-04 | 2017-10-20 | 华中科技大学 | A kind of character recognition method based on deep learning |
| CN108664996A (en) * | 2018-04-19 | 2018-10-16 | 厦门大学 | A kind of ancient writing recognition methods and system based on deep learning |
| CN108805223A (en) * | 2018-05-18 | 2018-11-13 | 厦门大学 | A kind of recognition methods of seal character text and system based on Incep-CapsNet networks |
| CN109376731A (en) * | 2018-08-24 | 2019-02-22 | 北京三快在线科技有限公司 | A kind of character recognition method and device |
| CN109492643A (en) * | 2018-10-11 | 2019-03-19 | 平安科技（深圳）有限公司 | Certificate recognition methods, device, computer equipment and storage medium based on OCR |
| CN109657739A (en) * | 2019-01-09 | 2019-04-19 | 西北大学 | A kind of hand-written Letter Identification Method based on high frequency sound wave Short Time Fourier Transform |
| CN109948696A (en) * | 2019-03-19 | 2019-06-28 | 上海七牛信息技术有限公司 | A kind of multilingual scene character recognition method and system |
-
2019
- 2019-10-24 CN CN201911015645.1A patent/CN110738188A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105335689A (en) * | 2014-08-06 | 2016-02-17 | 阿里巴巴集团控股有限公司 | Character recognition method and apparatus |
| CN104794455A (en) * | 2015-05-04 | 2015-07-22 | 北京信息科技大学 | Dongba hieroglyphic recognizing method |
| CN107273897A (en) * | 2017-07-04 | 2017-10-20 | 华中科技大学 | A kind of character recognition method based on deep learning |
| CN108664996A (en) * | 2018-04-19 | 2018-10-16 | 厦门大学 | A kind of ancient writing recognition methods and system based on deep learning |
| CN108805223A (en) * | 2018-05-18 | 2018-11-13 | 厦门大学 | A kind of recognition methods of seal character text and system based on Incep-CapsNet networks |
| CN109376731A (en) * | 2018-08-24 | 2019-02-22 | 北京三快在线科技有限公司 | A kind of character recognition method and device |
| CN109492643A (en) * | 2018-10-11 | 2019-03-19 | 平安科技（深圳）有限公司 | Certificate recognition methods, device, computer equipment and storage medium based on OCR |
| CN109657739A (en) * | 2019-01-09 | 2019-04-19 | 西北大学 | A kind of hand-written Letter Identification Method based on high frequency sound wave Short Time Fourier Transform |
| CN109948696A (en) * | 2019-03-19 | 2019-06-28 | 上海七牛信息技术有限公司 | A kind of multilingual scene character recognition method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7885464B2 (en) | Apparatus, method, and program for handwriting recognition | |
| EP2166488B1 (en) | Handwritten word spotter using synthesized typed queries | |
| US7729538B2 (en) | Spatial recognition and grouping of text and graphics | |
| Pratikakis et al. | ICFHR 2014 competition on handwritten keyword spotting (H-KWS 2014) | |
| CN111652332B (en) | Deep learning handwritten Chinese character recognition method and system based on two classifications | |
| JPH06243297A (en) | Method and equipment for automatic handwritten character recognition using static and dynamic parameter | |
| Fazilov et al. | State of the art of writer identification | |
| Valy et al. | Character and text recognition of khmer historical palm leaf manuscripts | |
| Kumar et al. | Unsupervised learning dimensionality reduction algorithm PCA for face recognition | |
| Ramesh et al. | Offline Kannada handwritten character recognition using convolutional neural networks | |
| Saabni et al. | Keyword searching for Arabic handwritten documents | |
| Liang et al. | Automatic handwriting feature extraction, analysis and visualization in the context of digital palaeography | |
| De Stefano et al. | Layout measures for writer identification in mediaeval documents | |
| Nadeem et al. | Character recognition using template matching | |
| Zhang et al. | Computational method for calligraphic style representation and classification | |
| CN110738188A (en) | Ancient character recognition system based on presorting | |
| Halder et al. | Individuality of isolated Bangla characters | |
| Joshi et al. | Combination of multiple image features along with KNN classifier for classification of Marathi Barakhadi | |
| Niels et al. | Introducing Trigraph-trimodal writer identification | |
| Halder et al. | Individuality of Bangla numerals | |
| Nirmalasari et al. | Handwritten Text Recognition using Fully Convolutional Network | |
| Khan et al. | Isolated handwritten pashto characters recognition using KNN classifier | |
| Wolf et al. | Improving handwritten word synthesis for annotation-free word spotting | |
| Gerhana et al. | Comparison of Template Matching Algorithm and Feature Extraction Algorithm in Sundanese Script Transliteration Application using Optical Character Recognition | |
| Elaiwat et al. | Arabic word recognition system for historical documents using multiscale representation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200131 |
|
| WD01 | Invention patent application deemed withdrawn after publication |